Machine learning predicts and optimizes hydrothermal liquefaction of biomass

Using evolutionary machine learning to characterize and optimize co-pyrolysis of biomass feedstocks and polymeric wastes

“…The GPR provided the best R 2 , MAE, and RMSE values in the training and testing phases. These findings could be because of the excellent capacity of the GPR to deal with noisy data and nonlinear systems (Jiang et al, 2021;Shafizadeh et al, 2022). The average R 2 scores of the GPR approach for pyrolysis oil, char, and syngas in the training phase were 0.960, 0.972, and 0.905, respectively.…”

“…To adjust weights and biases, the MLPNN model uses several activation functions (i.e., ReLU, sigmoid, and tanh functions). The capability of the MLPNN model to solve stochastic and complex problems makes it one of the most widely used ML models (Darvishan et al, 2018;Shafizadeh et al, 2022).…”

“…When calculating the velocity, each particle search for the optimum position according to its best previous position and the optimal position found by all particle. This iterative process of calculating velocity, updating position, and re-evaluating particles using the objective function is iteratively continued until the satisfaction of an end condition, which can be reaching a certain accuracy level or hitting a pre-determined maximum number of iterations (Shafizadeh et al, 2022). The PSO algorithm is able to solve problems with a singleobjective nature.…”

Turning hazardous volatile matter compounds into fuel by catalytic steam reforming: An evolutionary machine learning approach

“…The interpretable and flexible hybrid GAM model combines generalized linear and additive models to include nonparametric smoothing terms. In better words, GAM uses a linear combination of multiple smooth functions of explanatory variables to model the response of interest (Shafizadeh et al, 2022). Accordingly, this model not only can provide predicted values but also can help understand the effect of the co-variables on the response variables (Shafizadeh et al, 2022).…”

“…In better words, GAM uses a linear combination of multiple smooth functions of explanatory variables to model the response of interest (Shafizadeh et al, 2022). Accordingly, this model not only can provide predicted values but also can help understand the effect of the co-variables on the response variables (Shafizadeh et al, 2022). Generally, GAM can model complex problems without special conditions or unique solutions.…”